计算机工程与应用 ›› 2020, Vol. 56 ›› Issue (15): 179-184.DOI: 10.3778/j.issn.1002-8331.1906-0194

• 模式识别与人工智能 • 上一篇    下一篇

群智感知网络中基于社会关系的社区发现算法

龙浩,张书奎,张力   

  1. 1.苏州大学 计算机科学与技术学院,江苏 苏州 215006
    2.徐州工业职业技术学院 信息与电气工程学院,江苏 徐州 221002
    3.江苏省现代企业信息化应用支撑软件工程技术研发中心,江苏 苏州 215104
  • 出版日期:2020-08-01 发布日期:2020-07-30

Community Detection Algorithm Based on Social Relationship in Crowd Sensing Network

LONG Hao, ZHANG Shukui, ZHANG Li   

  1. 1.School of Computer Science and Technology, Soochow University, Suzhou, Jiangsu 215006, China
    2.School of Information and Electrical Engineering, Xuzhou College of Industrial Technology, Xuzhou, Jiangsu 221002, China
    3.Jiangsu Province Support Software Engineering R&D Center for Modern Information Technology Application in Enterprise, Suzhou, Jiangsu 215104, China
  • Online:2020-08-01 Published:2020-07-30

摘要:

群智感知应用主要通过社区划分进行任务分配,然而现有群智感知应用中社区发现算法缺乏对社会关系的量化以及划分社区的特征因子单一。针对这些问题,提出了一种基于多维社会关系特征的社区发现算法,通过计算移动节点间的最优生成树、节点合并因子、社区调整因子,对移动节点的社会关系进行具体量化,将节点合理划分成不同的社区。实验结果表明,与现有方法相比,该算法在不同的数据集中具有更好的动态适应性、有效性和预测准确性,其准确性平均达到97.3%,高于对比算法15.2%。

关键词: 群智感知, 特征因子, 最优生成树, 节点合并, 社区调整

Abstract:

The sensing tasks are mainly distributed through the community division in the crowd sensing application. However, the community discovery algorithm in the existing crowd sensing application lacks the quantification of social relations and the characteristic factors of the discovery community are single. In order to resolve these problems, a community detection algorithm based on multi-dimensional social relationship feature is proposed. By calculating the optimal spanning tree, the node merging factor and the community adjustment factor between mobile users, the social relations of the mobile nodes are specifically quantified and the users are divided into different communities. The experimental results show that compared with the existing methods, the algorithm has better dynamic adaptability, validity and prediction accuracy in different data sets and the accuracy is 97.3% on average, which is higher than the contrast algorithm by 15.2%.

Key words: crowd sensing, feature factor, optimal path tree, node merging, community adjustment